PARINITH
Hello, I'm

ParinithReddy

AI / ML Engineer·Backend Systems Builder

Building intelligent systems, scalable backends, and research-driven machine learning solutions.

View ProjectsDownload Resume
PythonPyTorchNext.jsFastAPIGoDocker
Scroll
01 / About

Who I Am

I'm a software developer studying Electronics & Communication Engineering at NIT Delhi, with a deep passion for the intersection of machine learning, systems design, and engineering.

I build AI/ML systems, scalable backends, and research-driven solutions - from real-time computer vision pipelines to distributed retrieval architectures.

Outside engineering, I lead design work at Upvision and contribute to my institute's digital infrastructure - bridging creative and technical thinking.

8.28
CGPA
175+
LeetCode
4+
Projects
2024NIT Delhi

Started B.Tech in Electronics & Communication Engineering. Maintained a CGPA of 8.28.

2025ISDC - NIT Delhi

Joined the Institute Software Development Cell. Contributed to the official NIT Delhi website (nitdelhi.ac.in) and backend systems.

2025Head Graphic Designer - Upvision

Led design and branding for Upvision Technical Club - events, standees, brochures, and promotional assets.

2026Research - LSH Image Search

Investigating Locality Sensitive Hashing for scalable approximate nearest neighbour image retrieval at NIT Delhi.

02 / Skills

Tech Stack

{ }

Languages

C++PythonGoJavaScriptTypeScriptSQL
⚙

Frameworks & Tools

ReactNext.jsFastAPIFlaskDockerAWSGitLinux
⬡

AI / ML

PyTorchTensorFlowScikit-learnOpenCVMediaPipeLangChain
◈

Concepts

DSAComputer VisionRAG SystemsDBMSDistributed Systems

Developer Tooling

GitLinuxDockerVS CodeJupyter NotebookFigmaVercelPostman
03 / Projects

What I've Built

Computer Vision

Handtracking Piano

Real-time computer vision piano controlled using hand tracking. Engineered using Google MediaPipe's pre-trained neural network for on-device hand landmark detection entirely client-side via WebAssembly.

  • 21 hand landmark detection across skeletal keypoints
  • Sub-50ms end-to-end inference latency
  • Gesture-to-audio pipeline with debounced collision detection
ReactTypeScriptMediaPipeTone.jsWebAssemblyVite
ML / Recommendation

Audify

Content-aware music recommendation engine trained on 10,000+ Spotify tracks using K-Means clustering over high-dimensional audio feature vectors — tempo, energy, valence, danceability.

  • K-Means over high-dimensional audio feature vectors
  • Hybrid: cluster-based retrieval + cosine similarity ranking
  • 80% intra-cluster retrieval, 20% cross-cluster diversity
PythonNext.jsScikit-learnSpotify APIK-Means
Full Stack · AI

PurePixels

Full-stack AI-powered background removal platform using deep learning segmentation. Scalable RESTful backend with PostgreSQL for auth, image metadata, and rate limiting across 100+ concurrent sessions.

  • Deep learning segmentation pipeline
  • Sub-5s processing latency under concurrent load
  • Per-user daily rate limiting across 100+ sessions
Next.jsTypeScriptFastAPIPostgreSQLTailwindCSS
ML · Data Science

F1 Strategy Prediction Engine

ML system predicting optimal Formula 1 pit strategies using FastF1 telemetry data from 200k+ laps across 20+ circuits and 4 seasons with chained multi-model architecture.

  • 200k+ lap telemetry dataset from 4 seasons
  • 3 chained ML models: XGBoost + Random Forest
  • Season-blocked cross-validation to prevent temporal leakage
PythonXGBoostPandasRandom ForestOptunaStreamlit
View all repositories on GitHub
04 / Research

Deep Dive

NIT Delhi - 2026 - PresentActive

Locality Sensitive Hashing for Image Similarity Search

Investigating approximate nearest neighbour (ANN) search techniques using Locality Sensitive Hashing (LSH) for scalable image retrieval over high-dimensional feature spaces. Implementing and benchmarking random projection-based hashing schemes on standard image datasets.

ANN Search Pipeline

Image Input→
Feature Embedding→
Random Projection
↓
Hash Buckets→
ANN Retrieval
Approximate Nearest Neighbours
Random Projections
Dimensionality Reduction
Hash Bucket Uniformity
Query Latency vs Recall Trade-offs
CIFAR-10 / MNIST Benchmarking

Research Interests

Scalable ML Algorithms
Dimensionality Reduction
Similarity Search
Locality Sensitive Hashing
Randomized Algorithms
Clustering with Provable Guarantees

Benchmarked Datasets

CIFAR-10

60,000 images

3072-D

MNIST

70,000 samples

784-D
05 / Achievements

Problem Solving

0+

Problems Solved

LeetCode

Algorithms · Database · CP

0+

Problems Combined

Codeforces + AtCoder + CodeChef

Competitive Programming

0+

DSA Problems

Total

Across all platforms

CGPA: 8.28 / 10

B.Tech ECE · National Institute of Technology, Delhi

DSAAlgorithmsDatabaseCompetitive
06 / Contact

Let's Connect

I'm open to internships, research collaborations, and full-time roles in AI/ML and backend engineering. Feel free to reach out.

Email

parinithreddymavurapu@gmail.com

Open

LinkedIn

linkedin.com/in/parinith-reddy

Open

GitHub

github.com/parinith-web

Open
New Delhi, India · +91 83285 34237